The analysis of species existing in a gas by virtue of their finite volatility is of interest in many situations, for instance, for detecting explosives or dangerous substances, in the food and aroma industries, in the identification of incipient symptoms of decomposition in foods, in medical diagnosis based on the composition of bodily fluids or breath, skin odors, etc. Because the species to be detected is in the gas phase, the dominant technique of such analyses has been gas chromatography coupled to mass spectrometry (GC-MS) [1]. However, the method is much slower and often less sensitive than the alternative of ionizing the vapors directly at atmospheric pressure and then introducing the resulting ions into a mass spectrometer with an atmospheric pressure source (API-MS). This approach was pioneered by the TAGA system developed at Sciex [2], where vapor ionization was achieved by means of an electrical discharge. A significant advance towards the development of detectors for trace gases was taken in U.S. Pat. No. 4,531,056 by J. Fenn and colleagues through their invention of so called electrospray mass spectrometry (ES-MS; see also reference [3]). This approach was not originally intended to apply to gases. However, Fenn and colleagues [4, 5, 6] noted that vapors put in contact with an electrospray cloud were efficiently ionized, with limits of detection in the parts per billion level (ppb=10−9 atmospheres of partial pressure). Earlier studies had already demonstrated excellent though inferior sensitivities for vapors based on ionizing them at atmospheric pressure and then analyzing them in instruments referred to as ion mobility spectrometers (IMS). In this case the ionization sources had been generally based on radioactive materials, such as Ni-63. But Wu et al. [7] had also obtained interesting results with an electrospray charger which they referred to as secondary electrospray ionization (SESI), which is, broadly speaking, analogous to that independently described by Fenn and colleagues (for an MS rather than an IMS analyzer). The relative merits of the corona discharge used in the TAGA instrument and the SESI charger have remained unstudied for a long time, probably for the same reasons that led to the interruption of the use of API-MS systems for volatile analysis. The status of this long dormant field has been recently reviewed in [8].
Other specialized schemes have been developed independently for volatile analysis involving alternative methods of charging vapors. One example is so-called proton transfer reactions (PTR), where the vapors are mixed with solvated protons in a fast flow at reduced pressure. Part per trillion (ppt=10−12 atmospheres of partial pressure) lowest detection limits have been reported, though only with vapors of relatively small molecular weight [9, 10].
Because the potential of API-MS analysis of volatiles is more easily achieved based on commercial API-MS instruments rather than specialized research instruments, we shall focus the subsequent discussion of prior art on the former type. The charging and sampling methods taught by Fenn and colleagues require some detail that will provide the background for later improvements. The electrospray mass spectrometry method they had introduced in U.S. Pat. No. 4,531,056 involves the use of a counterflow dry gas interposed between the atmospheric pressure inlet of the mass spectrometer and the electrospray source. Some typical elements of this system are shown in FIG. 1, together with other new features to be later discussed. The MS inlet (1) is most often a small orifice in a plate or the bore of a capillary, through which atmospheric gas is sampled at sonic speed into the vacuum system of the mass spectrometer (2). For the purpose of the present invention the analyzer is not necessarily a mass spectrometer, but could be similarly an IMS or a DMA. The counterflow gas, often nitrogen, bathes the region upstream of the sonic orifice (1), enclosed in a chamber open towards the atmosphere through a curtain plate orifice (3). Part of the counterflow gas is sampled into the vacuum system of the MS (2) through the orifice (1), forming a supersonic jet (4). The rest exits through the curtain plate orifice (3), forming a counterflow or curtain jet (5), initially coaxial with the sonic jet, but moving in the opposite direction towards the open atmosphere of the room. This counterflow gas is meant to avoid ingestion by the MS of condensable vapors or dust coming from either the electrospray drops or the surrounding atmosphere. Ions, however, are able to penetrate through the curtain gas, driven by electric fields against the counterflow. A similar approach in which the term curtain gas was first coined had been used in Sciex instruments prior to Fenn's work, with a different type of atmospheric pressure ionization source. Its origin can be traced back to U.S. Pat. No. 4,300,044 and the pioneering work if Iribarne and Thomson [11]. The counterflow gas used by Fenn and colleagues impinged frontally against the electrospray cloud (6), offering excellent contacting area between the dry gas and the charged drops and electrospray ions. This useful feature was used in [4, 5] for volatile charging to increase the vapor ionization probability by feeding controlled quantities of vapor mixed with the counterflow gas, thereby maximizing their contact with the charged cloud and hence the charging probability of the vapor species. Under these conditions they could report sensitivities “for some species at ppb levels or less” [5]. Although quite novel at the time, such sensitivities are unfortunately inadequate to detect explosives such as PETN or RDX. Another problem with this approach when used for the analysis of ambient species is that the sample ambient gas is generally not clean, whereby the mass spectrometer would be rapidly contaminated. Furthermore, condensation of ambient water vapor on the ions would seriously impair the operation of the MS (though this difficulty may be overcome in some cases by substantial heating of the sampled humid gas). One solution to sidestep this contamination problem is proposed in U.S. patent application Ser. No. 11/732,770 by Martinez-Lozano and Fernandez de la Mora, where the contaminated flow carrying the sample is fed into a chamber in which clean counterflow gas coming from the curtain plate orifice (3) flows directly against an electrospray cloud. This system contributes various improvements over prior art taught in [4, 5], whose combination has enabled record lowest detection levels as small as 0.2 ppt for trace vapor species [8], while also moderating the ingestion of dust, water vapor and other contaminants into the mass spectrometer. The setup of U.S. Ser. No. 11/732,770 is shown schematically in FIG. 2. Briefly, the vapors to be analyzed are ionized by contact with a source of charge, they are then drawn into a mass spectrometer in a fashion such that contaminant ingestion is greatly reduced. Finally, the transmission of ions into the analyzing section of the mass spectrometer is much enhanced by the use of so-called ion guides, as discussed for instance in U.S. Pat. No. 4,963,736, or in the related ion funnels of U.S. Pat. No. 6,107,628. Instead of carrying the vapors of interest to be analyzed (subsequently referred to as target vapors) with the counterflow, Martinez-Lozano and Fernandez de la Mora carry said vapors with another flow to be referred to as sample flow (7). In one single chamber (8), directly connected to the curtain plate of the mass spectrometer, they introduce the sample flow (7) laterally, while the ionization source (9) and the counterflow jet (5) are aligned along the same axis. In the preferred embodiment of U.S. Ser. No. 11/732,770, the ionization source is an ES source that produces the electrospray cloud (6).
Counterflow gas and dilution of the sample vapor in the ionization volume. In the publications making use of the charger of U.S. Ser. No. 11/732,770, the sample flow used was typically 6 lit/min, while the flow taken by the analyzer was only 0.5 lit/min [12, 8]. Although large with respect to the analyzer intake flow, these sample flow rates are in fact considerably smaller than those typical in the earlier TAGA system. But they are still relatively large for many applications.
In order to facilitate ionization of the sample and the ingestion of the resulting sample ions into the analyzer, the sample gas and the ionizing agents produced by the ionization source (9) must coexist in a volume where the streamlines formed by the velocity of the ions reach the entrance of the analyzer. This volume will be termed here the effective ionization volume. In the configuration of FIG. 2, where the ion source and the curtain plate orifice (3) are approximately coaxial, the ionization volume tends to be substantially occupied by clean counterflow gas. In order for the sample gas to be ionized, it must reach the effective ionization volume. This it can do either weakly by diffusion across the counterflow jet, or more vigorously by having sufficient momentum to deflect the counterflow jet (5) away from part of the effective ionization volume (as shown in FIG. 2). In this configuration, the ionization source (9) must be maintained at a certain distance from the curtain plate orifice (3), such that the counterflow jet (5) is sufficiently weakened to be deflected. The unbounded lateral impaction between the counterflow jet and the sample flow is typically unstable and leads to effective mixing between both flows. As a result, the vapors in the effective ionization volume are diluted by the counterflow.
The reasons why these substantial sample flows were previously needed to achieve good sensitivity have not been discussed in the published or patent literature. However, the sample flow rate clearly needs to be higher or at least of the same order as the counterflow to counteract dilution by the counterflow, and to partially deflect the counterflow jet away from the ionization volume. This notion can be expressed in terms of the dimensionless parameter to be referred to as the flow ratio q, defined as the ratio between the sample flow rate and the counterflow flow rate. Therefore, in the ionizer of U.S. Ser. No. 11/732,770, the flow ratio q has in principle to be of order unity or larger, and it is found in practice that it needs to be substantially larger. Under such conditions prior work [12] has achieved record high sensitivities, though at the cost (not always affordable) of consuming considerable sample flow.
The case of limited available sample. The need for relatively large q values in U.S. Ser. No. 11/732,770 does not appear to pose great problem, as long as the volume of gas to be analyzed is not substantially limited, such as when one samples from the open atmosphere or from a large room. However, in some applications, including explosive detection and skin vapor analysis, the rate at which the target species is incorporated into the gas sampled into the analyzer is limited. The total amount of the target species in the gas phase can also be limited if, for instance, it is desorbed from a collection or preconcentration device where target particles or vapors have been previously accumulated for a certain time period. In those cases, the concentration of vapors is inversely proportional to the sample flow rate and the scheme proposed by Martinez-Lozano is not able to efficiently use the limited available stock of sample. Having a high sample flow rate would inevitably dilute the sample with clean air before introducing it into the ionization chamber. And, if one tried to reduce the sample flow to avoid dilution at the source, the sample would still be highly diluted by the counterflow gas from the analyzer, while the region of coexistence between the target vapor and the ionization source would become small or could even disappear as the counterflow jet would occupy most of the effective ionization volume. Either using low sample flow rates or high flow rates therefore leads to high inefficiency.
The ionization probability and the target ion concentration. The behavior in the sample ionization region is peculiar when the ionization source is an electrospray or another ionization source producing preferentially ions of a single polarity. In this case, the rate at which vapor ionization takes place is proportional to the concentration nv of target vapors, the concentration nb of charger ions (to be so referred even though, as suggested by Fenn and colleagues, the charging agents may be electrospray drops), and a constant k governing the kinetics of the charge transfer reaction according to
                                                        Dn              i                        Dt                    =                                    kn              v                        ⁢                          n              b                                      ,                            (        1        )            where Dni/Dt is the production rate of target ions (ions per unit time and volume), and the concentrations nb and nv are expressed in units of molecules/volume. Provisionally, we presume that nv is undisturbed either by the counterflow and the ionization reaction itself, and will subsequently discuss how this can be achieved. The concentration of the charger ions is typically much higher than the concentration of target ions. As a result, the effect of target ions on the electric field can be neglected. On the other hand, the concentration of charge is proportional to the divergence of the electric field. Assuming stationary conditions, the net flow of target ions qi (ions/s) emanated from the ionization volume can be computed as the volume integral of the ionization rate through the effective ionization volume
                                          q            i                    =                                    ∫                              ∫                                  ∫                                                            kn                      v                                        ⁢                                          n                      b                                        ⁢                                          ⅆ                      V                                                                                            =                          ∫                              ∫                                  ∫                                                            kn                      v                                        ⁢                                                                  ɛ                        0                                            e                                        ⁢                                          ∇                                              ·                                                  E                          _                                                                                      ⁢                                          ⅆ                      V                                                                                                          ,                            (                  2          ,          a          ,          b                )            where we use Poisson's law, ∈0 is the permittivity of vacuum, e is the charge of an ion and E is the electric field.
Applying the Gauss theorem to the effective ionization volume and introducing the total velocity field composed by the electric velocity plus the fluid velocity, one can easily conclude that the net flow of target ions emanated from the ionization volume is equal to knv∈o/Zie (where Zi stands for the mobility of the target ions) times the flux of the electric and fluid velocities. Note that the second integral in (3), where Vf stands for the fluid velocity field, vanishes in the common circumstance in which the flow configuration is incompressible.
                              q          i                =                                                                              kn                  v                                ⁢                                  ɛ                  0                                                                              Z                  i                                ⁢                e                                      ⁡                          [                                                ∫                                      ∫                                                                                            (                                                                                                                    V                                _                                                            f                                                        +                                                                                          Z                                i                                                            ⁢                                                              E                                _                                                                                                              )                                                ·                                                  n                          _                                                                    ⁢                                              ⅆ                        A                                                                                            -                                  ∫                                      ∫                                                                                                                        V                            _                                                    f                                                ·                                                  n                          _                                                                    ⁢                                              ⅆ                        A                                                                                                        ]                                .                                    (        3        )            On the other hand, the net flow of target ions emanating from the ionization volume is:qi=∫∫ni( Vf+ZĒ)· ndA,  (4)Integrating both (3) and (4) through an infinitesimally thin stream tube, so that the concentration of ions can be considered constant along any section of the stream tube, the concentration of target ions in a section 1 compared to that of a section 2 is:
                                          n                          i              ⁢                                                          ⁢              2                                =                                                    n                v                            ⁢                                                                    k                    ⁢                                                                                  ⁢                                          ɛ                      0                                                                                                  Z                      i                                        ⁢                    e                                                  ·                                  (                                      1                    -                                                                  q                        1                                                                    q                        2                                                                              )                                                      +                                          n                                  i                  ⁢                                                                          ⁢                  1                                            ⁢                                                q                  1                                                  q                  2                                                                    ,                            (        5        )            where q1 and q2 stand for the infinitesimal flux of the velocity field through section 1 and 2 respectively. Note that ( Vf+ZĒ)· n=0 along the walls of the stream tube.
For the special case where the charger ions are created by means of an electrospray tip, the term q1/q2 tends to zero in the limit when the first section 1 of the stream tube is very close to the electrospray tip. Under these circumstances, the concentration of target ions is uniform and does not depend on the electrical or fluid configuration in the sample ionization region, but is simply given by
                                          n            i                    =                                    n              v                        ⁢                                          k                ⁢                                                                  ⁢                                  ɛ                  0                                                                              Z                  i                                ⁢                e                                                    ,                            (        6        )            This result was previously obtained by J. Fernandez de la Mora (Yale) for the case when the fluid velocity can be neglected compared with the electric velocity.
The case of an electrospray source is very specific because it has a singularity. In a more general case where the ion concentration does not tend to infinity in any region, the final concentration of target ions will be given by equation (5) and will be always lower than the limit expressed in equation (6). Nevertheless, the term q1/q2 can be reduced by means of the space charge effect as long as the amount of charger ions is significant enough.
The probability of ionization p has been previously defined [8] as the ratio between the concentration ni of sample ions carried to the analyzer and the maximum concentration theoretically available, which is the concentration nv of target vapors. According to equation (6), this probability of ionization p is independent of the sample flow rate:
                    p        =                                            n              i                                      n              v                                =                                                    k                ⁢                                                                  ⁢                                  ɛ                  0                                                                              Z                  i                                ⁢                e                                      .                                              (        7        )            The implications of this result are not altogether as good as one might hope from its elegant simplicity. The reason is that substitution of typical characteristic values for the various constants entering in equation (7) yield for atmospheric air: p˜10−4. But because this dismally low value is independent of essentially all the variables under control, one is apparently led to the conclusion that, of every vapor molecule available, only a rather small fraction p can be ionized, whose minute value is beyond our control. These unpleasant apparent conclusions are in fact overoptimistic, as they ignore the dilution effects due to the counterflow gas, as well as additional dilution (to be later analyzed) taking place as the target ions penetrate through the counterflow jet on their way to the mass spectrometer inlet. These discouraging theoretical estimates for p agree reasonably with the approximate measurements reported in [8].
The fact that the final concentration ni of target ions achievable is independent of flow rate is somewhat puzzling, and it is useful for the purposes of this invention to understand why. The rate equation (1) indicates that ni˜knvnbt, where t is a residence time. It follows that ni/nv˜knbt, which would normally increase with the residence time, and would ordinarily increase as the flow rate is decreased. However, this is not the case in our problem for two reasons. First, the time available for ionization is not determined by the fluid velocity, but, primarily, by the swifter electric drift velocity. As long as there is no counterflow dilution and p is small, the vapor concentration is relatively constant and equal to its source value. Consequently nv is a passive actor and it makes little difference on the final ni whether the neutral vapor is moving or not. In other words, the residence time of the neutral vapor is much larger than that of the ions moved by the field, and is therefore relatively irrelevant in the determination of ni. What really counts is the movement of the ions through the passive medium containing vapor molecules. Second, the concentration nb of charging ions is rapidly decreasing in time due to space charge. We shall subsequently see that, in the space charge controlled problem, the product nbt is in fact constant for an ion within the charged cloud, leading (in order of magnitude) to the same conclusion attained more rigorously in equation (6). This time can certainly be increased (by reducing the electric field or increasing the distance to be traveled from the tip of the ionizer to the analyzer). But then space charge decreases the concentration of charging ions, so that the effective nbt product is always the same. Space charge dilution is therefore the factor that limits p to the small and fixed values found when the charging ions are predominantly of only one polarity (unipolar ion source). This limitation has been previously recognized in PCT/EP2008/053960, where it was partially overcome by counteracting space charge repulsion with external radiofrequency fields.
In conclusion, prior attempts at ionizing vapors by interaction with charged drops and/or ions have encountered two kinds of limitations. First the serious dilution and expulsion effects of the target vapor away from the charging region in analyzers using counterflow gas. Second the tiny value of the maximum achievable charging probability resulting from the rapid space charge dilution of the charger ions. The first of these limitations is particularly harmful in circumstances when the sample available is limited.
Before proceeding to partially overcome these difficulties according to the present invention, it is instructive to introduce a charging probability more relevant than p in cases when the total quantity of sample gas available for analysis is limited. We define the single molecule probability of ionization pmi as the fraction of target gas molecules fed to the inlet of the ionizer that are transferred to the analyzer as ions. In the ideal case where counterflow dilution can be neglected, the probability of ionization and the single molecule probability of ionization are related as follows:
                                          p            mi                    =                      p            ⁢                                          Q                A                                            Q                s                                                    ,                            (        8        )            where QA is the flow rate of gas ingested by the analyzer; and QS is the flow rate of sample gas. This result shows clearly that when QA/QS>>1 one can apparently convert into ions a fraction of the neutral sample much larger than p. But how can this be done if ni/nv is fixed independently of QS?
In the answer to this question lies the key to one central aspects of the present invention. The sample is used at a rate QSnv. Yet, ni is fixed independently of QS. But the flux of target ions drawn into the analyzer is not necessarily QSni. It may in fact be much larger, as long as the electric drift velocity of these ions is much larger than typical flow velocities. In other words, space charge fixes the concentration of target ions, but not the flux at which they are extracted electrically. What one needs therefore to do is to increase this ion flux enough such that each parcel of gas sampled into the analyzer carries target ions at a concentration ni close to the value achievable in the charging chamber (in the absence of counterflow dilution). When QS is small compared to QA, but not so small as to make pmi of order unity (say pmi<0.1), the consumption of vapor molecules is small, and those ionized and removed by the field can easily be replaced by diffusion from those outside the charged plume. nv will hence remain comparable to its source value. Then equation (6) holds, and application of a suitable electric field will extract an adequate flux of target ions to feed them to the analyzer at concentration approaching ni. On the other hand, once QA/QS is large enough to make pmi of order unity, neutral vapors will be consumed fast enough for nv to be reduced below its source value, modifying the previous results so that pmi would never exceed unity, but would simply tend towards it. It is therefore possible in principle to approach the ideal limit when the majority of the sample vapor molecules are ionized and transmitted to the analyzer. The present invention aims at progressing towards this possible ideal within practical limits. In reality, of course, one would only have a finite time available to perform the analysis, so that QS would take a finite value. For example, suppose one wishes to analyze a sample of explosive molecules collected in a filter, where the volume of gas to be displaced from the filter into the analyzer is 5 cm3. Suppose further that the analysis is to be completed in 10 minutes, so that the sample flow rate would be of 0.5 cm3/min. If the flow rate into the analyzer is 0.5 lit/min, then QA/QS=103, whereby pmi would be 0.1 for p=10−4. This would imply a use of sample some 104 times more efficiently than in the work of [8] (where QA/QS˜0.1), which showed in turn a considerably greater sensitivity for vapor detection than any preceding study.
As just noted, when the sample flow is small, the ions have to be drawn from the charging region into the analyzer primarily by the electric field. However, this has not been done properly in any prior study. In U.S. Ser. No. 11/732,770, the principal means used to push the ions through the counterflow region is the electric field generated by the electrospray tip, which decays relatively fast with the distance to the tip. Furthermore, this tip must be placed relatively far from the analyzer inlet to avoid the effect of dilution produced by the counterflow. In one instance where the ionizer described in [8] could not be fitted into a desired quadrupole mass spectrometer analyzer (Sciex's API 5000), the sample gas was directly opposed to the counterflow gas, and a relatively weak auxiliary field besides that created by the electrospray needle was used. Neither of these approaches, however, provides an adequate control of the electric field to feed the entrance region of the analyzer with target ions at a concentration near the ideal value given in equation (6). As a result, even if dilution is avoided by some as yet undisclosed scheme, either many streamlines reaching the analyzer will carry clean gas without target ions at low sample flow rates, or the sample will be used inefficiently at high sample flow rates. The present invention will incorporate means to apply the necessary fields to fill most streamlines entering the analyzer with ions at a concentration close to the ideal value of equation (6).
In conclusion, prior studies have succeeded at moderating the dilution associated to counterflow gas only at the cost of using high sample flow rates. In situations where the finite sample available must be used efficiently, whereby QS/QA needs to be small, no solution has been available to either avoid sample dilution due to counterflow gas, or to drive the target ions efficiently into the analyzer inlet. Consequently, the purposes of the present invention are to teach                (i) How to prevent dilution of neutral target vapors in the ionization region due to counterflow gas, and thus maximize the concentration of the sample flow in the ionization region;        (ii) How to fill with target ions the majority of the fluid streamlines sucked into the analyzer, and how to minimize the dilution of target ions due to diffusion and space charge effects as they cross a clean counterflow region.        (iii) How to reduce drastically the required sample flow, even in the presence of counterflow, and thus how to increase the single molecule probability of ionization while minimizing the dilution effects due to the counterflow.        (iv) How to reduce the flow of charger ions qb ingested by the analyzer without reducing the flow of target ions        